A Computer Vision Approach to Build Manufacturing Simulation Models in the Digital Twin Context
摘要
The digital transformation with the application of immersive mixed reality has been supporting the digital twin creation within the industrial context. The techniques for 3D reconstruction from the Computer Vision approach make possible the creation of virtual models of real scenery based on a set of digital image frames. In this paper, we explored the problem of building virtual layouts from 3D mapping by photogrammetry to be applied in the corresponding digital twin model. The context of Industry 5.0 is aborded in the research and aims to define a method for preparing humans to be trained using synthetic information, to perform the real operations in the productive processes. From collected RGB images, the layout of a shopfloor can be recreated, and objects can be virtually represented. However, to rebuild a realistic environment is a difficult task. Thus, we studied the main technological requirements to perform an automated method for shop-floor layout digitalization and digital twin creation, conducting an experimental investigation for a manufacturing automation laboratory. The work also discusses the feasibility of the proposed method.